基于P2P-CDN的网络自适应分片及调度

发布时间：2018-10-18 12:24

【摘要】：基于对等网络(P2P)的内容请求由于其灵活性、分发负载小、有效性高等众多优点而被广泛使用。但是,由于网络和系统环境变化,仍存在一些问题,如服务有效性、稳定性和自适应性等。这些问题的解决,可以进一步提高系统性能和服务质量,改善用户的观看体验。当前P2P网络系统,主要分为有结构、无结构和混合型三种。其中混合型兼具结构和无结构模型优点,查询有效性高且系统稳定性强。本文基于混合型P2P网络,并结合内容分发策略设计了一种流媒体服务系统结构,通过设立局部对等网络自治域以提高系统服务性能,减少骨干网的流量负载。在基于P2P网络的系统中,为适合网络传输,对内容进行分片。常用的内容分片算法有等大小算法、等时长算法和变大小算法,它们在不同的网络环境下性能差异较大,并且不能适应网络变化。为此,本文提出一种自适应分片算法,来增强系统服务稳定性。该算法根据网络传输速率、丢包率、文件分布和不同节点的处理性能,来设计分片大小。实验结果表明,在大多数网络环境下,这种自适应分片算法比其它分片算法性能更好。现有分片调度算法有基于时间的顺序调度、稀有分片优先调度和基于权值的分片调度等,这些算法没有综合考虑视频的播放紧急程度、稀缺分片、用户与视频交互等因素。为了进一步提高系统服务性能,本文提出一种基于标签优先级的分片调度算法,根据视频当前播放时间点,将需要下载的分片根据优先级顺序打上不同标签:紧急分片、锚点分片、稀缺分片、普通分片和失效分片。当优先级一致,则根据时间先后来排序。实验结果表明,相对于基于时间顺序、稀缺分片优先和基于权值的调度算法,该算法性能提高显著。最后,在安卓系统的机顶盒上,对该系统中核心的内容分片和分片调度模块进行了验证。
[Abstract]:Peer-to-Peer (P2P) based content requests are widely used because of their flexibility, low distribution load and high efficiency. However, due to the network and system environment changes, there are still some problems, such as service effectiveness, stability and adaptability. The solution of these problems can further improve the system performance and service quality, and improve the user's viewing experience. At present, P2P network system is mainly divided into three types: structured, unstructured and hybrid. The hybrid model has the advantages of both structural and unstructured models, high query efficiency and strong system stability. This paper designs a streaming media service system structure based on hybrid P2P network and content distribution strategy. By setting up local peer-to-peer network autonomous domain to improve the system service performance and reduce the traffic load of backbone network. In the system based on P2P network, the content is partitioned to fit the network transmission. The commonly used content slicing algorithms include equal size algorithm, equal time length algorithm and variable size algorithm, which have different performance in different network environments, and can not adapt to network changes. Therefore, an adaptive slicing algorithm is proposed to enhance the system service stability. The chip size is designed according to the network transmission rate, packet loss rate, file distribution and processing performance of different nodes. The experimental results show that this adaptive slicing algorithm performs better than other algorithms in most network environments. The existing piecewise scheduling algorithms include sequential scheduling based on time, rare piecewise priority scheduling and weighted piecewise scheduling. These algorithms do not take into account the emergency degree of video playback, the scarcity of slicing, the interaction between users and video, and so on. In order to further improve the performance of the system, this paper proposes a slicing scheduling algorithm based on label priority. According to the current playback time point of the video, the segments that need to be downloaded are tagged according to the priority order: emergency slicing. Anchor points, scarce, ordinary, and failure segments. When the priority is the same, it is sorted according to the order of time. The experimental results show that the performance of the proposed algorithm is significantly improved compared with the scheduling algorithm based on the time sequence, the scarcity slicing priority and the weight based algorithm. Finally, in the Android set-top box, the core of the system content slicing and partitioning scheduling module is verified.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.02

【参考文献】